The purpose of this study was the descriptive analysis of patients with systemic lupus erythematosus (SLE) with a particular focus on initial clinical features, evolution and outcome of disease, prevalence of clinical and serological manifestations and identification of clinicoserological associations indicative of renal and CNS involvement. The methodology applied was the following: retrospective analysis of the clinical charts of 292 unselected patients (246 female (84.2%) and 46 male (15.7%)) with SLE examined between 1982 and 1992. Multivariate analysis and hierarchical log linear models were used to examine for clinicoserological associations. Descriptive analysis was based on the prevalence of main clinicoserological features and disease outcome. The outcome was examined on the basis of the number of flares, the presence of chronic renal failure, the presence of central nervous system (CNS) involvement with subsequent disability and deaths. Flares were considered the severe alterations in disease status, requiring additional therapy to be controlled. The disease begins most frequently in the second and third decade of life with cutaneous and joint manifestations, while renal and CNS involvement developed later. The prevalence of serious renal, pulmonary and CNS involvement as well as the prevalence of RF, anti-Sm and anti-nRNP antibodies remain low. Multivariate analysis revealed the associations of renal involvement with leukopenia and serositis, of anti-Sm with leukopenia, of secondary Sjogren's syndrome with RF and of thromboembolic events with anticardiolipin antibodies. Patients with childhood onset SLE have a higher tendency for developing renal involvement than adult onset SLE patients. In addition, anti-Ro(SSA) antibodies were associated with anti-La(SSB) and RF, while anti-Sm antibodies were associated with anti-nRNP and RF. Flares occurred with a frequency of 0.07 per patient per year. Only 63.6% of flares were accompanied by positive anti-dsDNA reactivities. Reported deaths were 0.0047 per patient per year. Hierarchical log linear models indicated that the main variables of the disease were sufficient to describe our disease model and that the order of the interaction between the variables was insignificant. We conclude that the prevalence of various clinical features associated with SLE is similar, although the prevalence of CNS and pulmonary involvement as well as anti-Sm and anti-nRNP antibodies are less prominent in Greek SLE patients than that reported in the literature. The various clinicoserological associations detected do not appear to be of major significance as they are not powerful enough to subgroup the disease.